Connectivity and performance tradeoffs in the cascade correlationlearning architecture

The cascade correlation is a very flexible, efficient and fast algorithm for supervised learning. It incrementally builds the network by adding hidden units one at a time, until the desired input/output mapping is achieved. It connects all the previously installed units to the new unit being added. Consequently, each new unit in effect adds a new layer and the fan-in of the hidden and output units keeps on increasing as more units get added. The resulting structure could be hard to implement in VLSI, because the connections are irregular and the fan-in is unbounded. Moreover, the depth or the propagation delay through the resulting network is directly proportional to the number of units and can be excessive. We have modified the algorithm to generate networks with restricted fan-in and small depth (propagation delay) by controlling the connectivity. Our results reveal that there is a tradeoff between connectivity and other performance attributes like depth, total number of independent parameters, and learning time.     

COVIDScreen: explainable deep learning framework for differential diagnosis of COVID-19 using chest X-rays

Neural Computing and Applications - COVID-19 has emerged as a global crisis with unprecedented socio-economic challenges, jeopardizing our lives and livelihoods for years to come. The...     

Camera distortion self-calibration using the plumb-line constraint and minimal Hough entropy

In this paper, we present a simple and robust method for self-correction of camera distortion using single images of scenes which contain straight lines. Since the most common distortion can be modelled as radial distortion, we illustrate the method using the Harris radial distortion model, but the method is applicable to any distortion model. The method is based on transforming the edgels of the distorted image to a 1-D angular Hough space, and optimizing the distortion correction parameters which minimize the entropy of the corresponding normalized histogram. Properly corrected imagery will have fewer curved lines, and therefore less spread in Hough space. Since the method does not rely on any image structure beyond the existence of edgels sharing some common orientations and does not use edge fitting, it is applicable to a wide variety of image types. For instance, it can be applied equally well to images of texture with weak but dominant orientations, or images with strong vanishing points. Finally, the method is performed on both synthetic and real data revealing that it is particularly robust to noise.     

Modeling of laser cladding with powder injection

Laser cladding is one of the material additive manufacturing processes used to produce a metallurgically bonded deposition layer. To obtain a high-quality resulting part, a deep understanding of the underlying mechanisms is required. In this article, a mathematical model is developed to simulate the coaxial laser-cladding process with powder injection, which includes laser- substrate, laser-powder, and powder-substrate interactions. The model considers most of the associated phenomena, such as melting, solidification, evaporation, evolution of the free surface, and powder injection. The fluid flow in the melt pool, which is mainly driven by Marangoni shear stress as well as particle impinging, together with the energy balances at the liquid-vapor and the solid-liquid interfaces, are investigated. Powder heating and laser power attenuation due to the powder cloud are incorporated into the model in the calculation of the temperature distribution. The influences of the powder injection on the melt pool shape, penetration, and flow pattern are predicted through the comparison for the cases with powder injection and without powder injection. Dynamic behavior of the melt pool and the formation of the clad are simulated. The effects of the process parameters on the melt pool dimension and peak temperature are further investigated based on the validated model.     

Synthesis of submicron size silver powder for passive components application

Submicron silver powder was prepared from AgNO₃ by a chemical reduction method in the presence of a mixture of caprylic acid and triethanolamine as a surfactant. Hydrazine hydrate (N₂H₄·H₂O) is preferred as a reducing agent. A spherical silver powder with an average particle size of about 150 nm was achieved. Effort was also made to correlate the crystal structure and microstructure evolution of the prepared powders with the resultant thick film characteristics.